Ensembling fine-tuned models initialized from powerful pre-trained weights is a common strategy to improve robustness under distribution shifts, but it comes with substantial computational costs due to the need to train and store multiple models. Dropout offers a lightweight alternative by simulating ensembles through random neuron deactivation; however, when applied to pre-trained models, it tends to over-regularize and disrupt critical representations necessary for generalization. In this work, we investigate Mixout, a stochastic regularization technique that provides an alternative to Dropout for domain generalization. Rather than deactivating neurons, Mixout mitigates overfitting by probabilistically swapping a subset of fine-tuned weights with their pre-trained counterparts during training, thereby maintaining a balance between adaptation and retention of prior knowledge. Our study reveals that achieving strong performance with Mixout on domain generalization benchmarks requires a notably high masking probability of 0.9 for ViTs and 0.8 for ResNets. While this may seem like a simple adjustment, it yields two key advantages for domain generalization: (1) higher masking rates more strongly penalize deviations from the pre-trained parameters, promoting better generalization to unseen domains; and (2) high-rate masking substantially reduces computational overhead, cutting gradient computation by up to 45% and gradient memory usage by up to 90%. Experiments across five domain generalization benchmarks—PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet—using ResNet and ViT architectures show that our approach, High-rate Mixout, achieves out-of-domain accuracy comparable to ensemble-based methods while significantly reducing training costs.

Object detection (OD) in infrared (IR) imagery is critical for low-light and nighttime applications. However, the scarcity of large-scale IR datasets forces models to rely on weights pre-trained on RGB images. While fine-tuning on IR improves accuracy, it often compromises robustness under distribution shifts due to the inherent modality gap between RGB and IR. To address this, we introduce LLVIP-C and FLIR-C, two cross-modality out-of-distribution (OOD) benchmarks built by applying corruptions to standard IR datasets. Additionally, to fully leverage the complementary knowledge from RGB and infrared-trained models, we propose WiSE-OD, a weight-space ensembling method with two variants: WiSE-ODZS, which combines RGB zero-shot and IR fine-tuned weights, and WiSE-ODLP, which blends zero-shot and linear probing. Evaluated using four RGB-pretrained detectors and two robust baselines on our benchmark and in the real-world out-of-distribution M3FD dataset, our WiSE-OD improves robustness across modalities and to corruption in synthetic and real-world distribution shifts without any additional training or inference costs. Our code is available at: https://github.com/heitorrapela/wiseod.

Robust visual recognition on embedded platforms requires models that both generalize out-of-distribution (OOD) and fit into tiny compute/memory budgets. While pre-training is a standard route to robustness for mid/large backbones, its value in the ultra-small regime remains unclear. We present a capacity-aware study of pre-training for two efficient ConvNet families (EfficientNet and MobileNetV3) scaled from "small" to "ultra-small" via a simple, reproducible recipe. We compare three initializations — ImageNet→COCO pretraining, ImageNet classification pretraining, and training from scratch—on two axes of distribution shift: (i) cross-dataset RGB→RGB transfer between LLVIP and FLIR (ii) cross-modality detection where models are fine-tuned on RGB and evaluated on infrared (IR). A complementary classification study on DomainNet probes whether the trends extend beyond detection. Across settings, we find that pretraining’s benefit is conditional on both backbone capacity and shift difficulty. Task-aligned Imagenet→COCO pretraining is the most reliable starting point at moderate sizes and for the easier transfer direction. In the low-capacity regimes, differences are typically within run-to-run variation, and training from scratch can match or surpass pre-training. Classification mirrors this capacity gating. Our results test the premise "pretraining always helps" and instead quantify when task-aligned pretraining pays off for ultra-small backbones and when it likely does not1.

Adapting person re-identification (reID) models to new target environments remains a challenging problem that is typically addressed using unsupervised domain adaptation (UDA) methods. Recent works show that when labeled data originates from several distinct sources (e.g., datasets and cameras), considering each source separately and applying multi-source domain adaptation (MSDA) typically yields higher accuracy and robustness compared to blending the sources and performing conventional UDA. However, state-of-the-art MSDA methods learn domain-specific backbone models or require access to source domain data during adaptation, resulting in significant growth in training parameters and computational cost. In this paper, a Source-free Adaptive Gated Experts (SAGE-reID) method is introduced for person reID. Our SAGE-reID is a cost-effective, source-free MSDA method that first trains individual source-specific low-rank adapters (LoRA) through source-free UDA. Then, a lightweight gating network is introduced and trained to dynamically assign optimal merging weights for fusion of LoRA experts, enabling effective cross-domain knowledge transfer. While the number of backbone parameters remains constant across source domains, LoRA experts scale linearly but remain negligible in size (ď 2% per source), reducing both the memory consumption and risk of overfitting. Extensive experiments conducted on three challenging benchmarks – Market-1501, DukeMTMC-reID, and MSMT17 – indicate that SAGE-reID can outperform state-of-the-art methods while remaining computationally efficient. Our code is available: https://github.com/nehdiii/SAGE-reID

Personalized expression recognition (ER) involves adapting a machine learning model to subject-specific data for improved recognition of expressions with considerable inter-personal variability. Subject-specific ER can benefit significantly from multi-source domain adaptation (MSDA) methods – where each domain corresponds to a specific subject – to improve model accuracy and robustness. Despite promising results, state-of-the-art MSDA approaches often overlook multimodal information or blend sources into a single domain, limiting subject diversity and failing to explicitly capture unique subject-specific characteristics. To address these limitations, we introduce MuSACo, a multi-modal subject-specific selection and adaptation method for ER based on co-training. It leverages complementary information across multiple modalities and multiple source domains for subject-specific adaptation. This makes MuSACo particularly relevant for affective computing applications in digital health, such as patient-specific assessment for stress or pain, where subject-level nuances are crucial. MuSACo selects source subjects relevant to the target and generates pseudo-labels using the dominant modality for class-aware learning, in conjunction with a class-agnostic loss to learn from less confident target samples. Finally, source features from each modality are aligned, while only confident target features are combined. Experimental results on challenging multimodal ER datasets – BioVid, StressID, and BAH – show that MuSACo outperforms UDA (blending) and state-of-the-art MSDA methods. Our code is available: https://github.com/osamazeeshan/MuSACo

The learning using privileged information paradigm leverages relevant features unavailable at deployment time for model training. In this paper, we propose a multi-task privileged framework that combines two types of tasks. First, the privileged-prediction task involves using regular features (available in both training and deployment) to predict privileged information, working as an intermediate step to guide the learning process. Second, the main learning objective, the target task, uses the predicted privileged information along with the regular features to make the final target prediction. Furthermore, knowledge distillation techniques are included within the target task to enhance the knowledge transfer of privileged information. Experimental results show improvements in tabular datasets and image-related problems compared to state-of-the-art approaches. Additionally, we analyze misclassification causes and refine the proposed multi-task privileged learning to reduce errors.

Personalized facial expression recognition (FER) involves adapting a machine learning model using samples from labeled sources and unlabeled target domains. Given the challenges of recognizing subtle expressions with considerable interpersonal variability, state-of-the-art unsupervised domain adaptation (UDA) methods focus on the multi-source UDA (MSDA) setting, where each domain corresponds to a specific subject, and improve model accuracy and robustness. However, when adapting to a specific target, the diverse nature of multiple source domains translates to a large shift between source and target data. State-of-the-art MSDA methods for FER address this domain shift by considering all the sources to adapt to the target representations. Nevertheless, adapting to a target subject presents significant challenges due to large distributional differences between source and target domains, often resulting in negative transfer. In addition, integrating all sources simultaneously increases computational costs and causes misalignment with the target. To address these issues, we propose a progressive MSDA approach that gradually introduces information from subjects (source domains) based on their similarity to the target subject. This will ensure that only the most relevant sources from the target are selected, which helps avoid the negative transfer caused by dissimilar sources. During adaptation, the source domains are introduced in a curriculum manner. We first exploit the closest sources to reduce the distribution shift with the target and then move towards the furthest while only considering the most relevant sources based on the predetermined threshold. Furthermore, to mitigate catastrophic forgetting caused by the incremental introduction of source subjects, we implemented a density-based memory mechanism that preserves the most relevant historical source samples for adaptation. Our extensive experiments 1 show the effectiveness of our proposed method on challenging FER datasets: Biovid, UNBC-McMaster, Aff-Wild2, and BAH. Further, performance is evaluated on a cross-dataset setting (UNBC-McMaster → BioVid), showing the importance of gradually adapting to source subjects.

Scalable Vector Graphics (SVG) have become integral to modern image rendering applications due to their infinite scalability and versatility, especially in graphic design and web development. SVGs are essentially long strings of code that adhere to a structured syntax with validity constraints. With the rise of large language models, which excel at generating code in various languages, we aim to generate SVG code in a similar way. Our findings show that a vision-language model can be conditioned to produce valid SVG code that closely resembles input images, effectively enabling vectorization. Additionally, we harness the rich SVG syntax, encompassing all possible primitives—such as lines, paths, polygons, text, and effects like color gradients—that previous methods often missed. We briefly explain how the StarVector model operates, primarily leveraging a vision-language transformer architecture to generate SVG code. We also detail our training and inference procedures. Finally, we provide an interactive demo that allows users to input an image and generate its SVG code autoregressively, featuring real-time rendering that visually demonstrates the SVG generation process.

Scalable Vector Graphics (SVGs) are vital for modern image rendering due to their scalability and versatility. Previous SVG generation methods have focused on curve-based vectorization, lacking semantic understanding, often producing artifacts, and struggling with SVG primitives beyond path curves. To address these issues, we introduce StarVector, a multimodal large language model for SVG generation. It performs image vectorization by understanding image semantics and using SVG primitives for compact, precise outputs. Unlike traditional methods, StarVector works directly in the SVG code space, leveraging visual understanding to apply accurate SVG primitives. To train StarVector, we create SVG-Stack, a diverse dataset of 2M samples that enables generalization across vectorization tasks and precise use of primitives like ellipses, polygons, and text. We address challenges in SVG evaluation, showing that pixel-based metrics like MSE fail to capture the unique qualities of vector graphics. We introduce SVG-Bench, a benchmark across 10 datasets, and 3 tasks: Image-to-SVG, Text-to-SVG generation, and diagram generation. Using this setup, StarVector achieves state-of-the-art performance, producing more compact and semantically rich SVGs.

Facial Expression Recognition (FER) from videos is a crucial task in various application areas, such as human-computer interaction and health diagnosis and monitoring (e.g., assessing pain and depression). Beyond the challenges of recognizing subtle emotional or health states, the effectiveness of deep FER models is often hindered by the considerable inter-subject variability in expressions. Source-free (unsupervised) domain adaptation (SFDA) methods may be employed to adapt a pre-trained source model using only unlabeled target domain data, thereby avoiding data privacy, storage, and transmission issues. Typically, SFDA methods adapt to a target domain dataset corresponding to an entire population and assume it includes data from all recognition classes. However, collecting such comprehensive target data can be difficult or even impossible for FER in healthcare applications. In many real-world scenarios, it may be feasible to collect a short neutral control video (which displays only neutral expressions) from target subjects before deployment. These videos can be used to adapt a model to better handle the variability of expressions among subjects. This paper introduces the Disentangled SFDA (DSFDA) method to address the challenge posed by adapting models with missing target expression data. DSFDA leverages data from a neutral target control video for end-to-end generation and adaptation of target data with missing non-neutral data. Our method learns to disentangle features related to expressions and identity while generating the missing non-neutral expression data for the target subject, thereby enhancing model accuracy. Additionally, our self-supervision strategy improves model adaptation by reconstructing target images that maintain the same identity and source expression. Experimental results1on the challenging BioVid, UNBC-McMaster and StressID datasets indicate that our DSFDA approach can outperform state-of-the-art adaptation methods.1https://github.com/MasoumehSharafi/DSFDA/

The zero-shot performance of object detectors degrades when tested on different modalities, such as infrared and depth. While recent work has explored image translation techniques to adapt detectors to new modalities, these methods are limited to a single modality and apply only to traditional detectors. Recently, vision-language detectors, such as YOLO-World and Grounding DINO, have shown promising zero-shot capabilities, however, they have not yet been adapted for other visual modalities. Traditional fine-tuning approaches compromise the zero-shot capabilities of the detectors. The visual prompt strategies commonly used for classification with vision-language models apply the same linear prompt translation to each image, making them less effective. To address these limitations, we propose ModPrompt, a visual prompt strategy to adapt vision-language detectors to new modalities without degrading zero-shot performance. In particular, an encoder-decoder visual prompt strategy is proposed, further enhanced by the integration of inference-friendly modality prompt decoupled residual, facilitating a more robust adaptation. Empirical benchmarking results show our method for modality adaptation on two vision-language detectors, YOLO-World and Grounding DINO, and on challenging infrared (LLVIP, FLIR) and depth (NYUv2) datasets, achieving performance comparable to full fine-tuning while preserving the model's zero-shot capability. Code available at: https://github.com/heitorrapela/ModPrompt.

Real-world datasets typically exhibit long-tailed (LT) distributions, where a few head classes dominate and many tail classes are severely underrepresented. While recent work shows that parameter-efficient fine-tuning (PEFT) methods like LoRA and AdaptFormer preserve tail-class performance on foundation models such as CLIP, we find that they do so at the cost of head-class accuracy. We identify the head-tail ratio, the proportion of head to tail classes, as a crucial but overlooked factor influencing this trade-off. Through controlled experiments on CIFAR100 with varying imbalance ratio ($\rho$) and head-tail ratio ($\eta$), we show that PEFT excels in tail-heavy scenarios but degrades in more balanced and head-heavy distributions. To overcome these limitations, we propose LT-Soups, a two-stage model soups framework designed to generalize across diverse LT regimes. In the first stage, LT-Soups averages models fine-tuned on balanced subsets to reduce head-class bias; in the second, it fine-tunes only the classifier on the full dataset to restore head-class accuracy. Experiments across six benchmark datasets show that LT-Soups achieves superior trade-offs compared to both PEFT and traditional model soups across a wide range of imbalance regimes.

Aligning visual features with language embeddings is a key challenge in vision-language models (VLMs). The performance of such models hinges on having a good connector that maps visual features generated by a vision encoder to a shared embedding space with the LLM while preserving semantic similarity. Existing connectors, such as multilayer perceptrons (MLPs), lack inductive bias to constrain visual features within the linguistic structure of the LLM’s embedding space, making them data-hungry and prone to cross-modal misalignment. In this work, we propose a novel vision-text alignment method, AlignVLM, that maps visual features to a weighted average of LLM text embeddings. Our approach leverages the linguistic priors encoded by the LLM to ensure that visual features are mapped to regions of the space that the LLM can effectively interpret. AlignVLM is particularly effective for document understanding tasks, where visual and textual modalities are highly correlated. Our extensive experiments show that AlignVLM achieves state-of-the-art performance compared to prior alignment methods, with larger gains on document understanding and under low-resource setups. We provide further analysis demonstrating its efficiency and robustness to noise.

Scalable Vector Graphics (SVG) offer a powerful format for representing visual designs as interpretable code. Recent advances in vision-language models (VLMs) have enabled high-quality SVG generation by framing the problem as a code generation task and leveraging large-scale pretraining. VLMs are particularly suitable for this task as they capture both global semantics and fine-grained visual patterns, while transferring knowledge across vision, natural language, and code domains. However, existing VLM approaches often struggle to produce faithful and efficient SVGs because they never observe the rendered images during training. Although differentiable rendering for autoregressive SVG code generation remains unavailable, rendered outputs can still be compared to original inputs, enabling evaluative feedback suitable for reinforcement learning (RL). We introduce Reinforcement Learning from Rendering Feedback, an RL method that enhances SVG generation in autoregressive VLMs by leveraging feedback from rendered SVG outputs. Given an input image, the model generates SVG roll-outs that are rendered and compared to the original image to compute a reward. This visual fidelity feedback guides the model toward producing more accurate, efficient, and semantically coherent SVGs. \method significantly outperforms supervised fine-tuning, addressing common failure modes and enabling precise, high-quality SVG generation with strong structural understanding and generalization.

Domain adaptation methods for object detection (OD) strive to mitigate the impact of distribution shifts by promoting feature alignment across source and target domains. Multi-source domain adaptation (MSDA) allows leveraging multiple annotated source datasets and unlabeled target data to improve the accuracy and robustness of the detection model. Most state-of-the-art MSDA methods for OD perform feature alignment in a class-agnostic manner. This is challenging since the objects have unique modality information due to variations in object appearance across domains. A recent prototype-based approach proposed a class-wise alignment, yet it suffers from error accumulation caused by noisy pseudo-labels that can negatively affect adaptation with imbalanced data. To overcome these limitations, we propose an attention-based class-conditioned alignment method for MSDA, designed to align instances of each object category across domains. In particular, an attention module combined with an adversarial domain classifier allows learning domain-invariant and class-specific instance representations. Experimental results on multiple benchmarking MSDA datasets indicate that our method outperforms state-of-the-art methods and exhibits robustness to class imbalance, achieved through a conceptually simple class-conditioning strategy. Our code is available at: https://github.com/imatif17/ACIA.

In real-world scenarios, using multiple modalities like visible (RGB) and infrared (IR) can greatly improve the performance of a predictive task such as object detection (OD). Multimodal learning is a common way to leverage these modalities, where multiple modality-specific encoders and a fusion module are used to improve performance. In this paper, we tackle a different way to employ RGB and IR modalities, where only one modality or the other is observed by a single shared vision encoder. This realistic setting requires a lower memory footprint and is more suitable for applications such as autonomous driving and surveillance, which commonly rely on RGB and IR data. However, when learning a single encoder on multiple modalities, one modality can dominate the other, producing un-even recognition results. This work investigates how to efficiently leverage RGB and IR modalities to train a common transformer-based OD vision encoder while countering the effects of modality imbalance. For this, we introduce a novel training technique to Mix Patches (MiPa)from the two modalities, in conjunction with a patch-wise modality agnostic module, for learning a common representation of both modalities. Our experiments show that MiPa can learn a representation to reach competitive results on traditional RGB/IR benchmarks while only requiring a single modality during inference. Our code is available at: https://github.com/heitorrapela/MiPa.

Weakly Supervised Object Localization (WSOL) allows training deep learning models for classification and localization (LOC) using only global class-level labels. The absence of bounding box (bbox) supervision during training raises challenges in the literature for hyper-parameter tuning, model selection, and evaluation. WSOL methods rely on a validation set with bbox annotations for model selection, and a test set with bbox annotations for threshold estimation for producing bboxes from localization maps. This approach, however, is not aligned with the WSOL setting as these annotations are typically unavailable in real-world scenarios. Our initial empirical analysis shows a significant decline in LOC performance when model selection and threshold estimation rely solely on class labels and the image itself, respectively, compared to using manual bbox annotations. This highlights the importance of incorporating bbox labels for optimal model performance. In this paper11Our scope. This work focuses on WSOL [2],[28]–[30], as opposed to other related tasks such as weakly supervised detection, segmentation, or instance segmentation, which are often mixed in earlier works [11], [18], [40]., a new WSOL evaluation protocol is proposed that provides LOC information without the need for manual bbox annotations. In particular, we generated noisy pseudo-boxes from a pretrained off-the-shelf region proposal method such as Selective Search, CLIP, and RPN for model selection. These bboxes are also employed to estimate the threshold from LOC maps, circumventing the need for test-set bbox annotations. Our experiments22Our code and generated pseudo-bounding boxes can be accessed at github.com/shakeebmurtaza/wsol_model_selection. with several WSOL methods on challenging natural and medical image datasets show that using the proposed pseudo-bboxes for validation facilitates the model selection and threshold estimation, with LOC performance comparable to models selected using GT bboxes on the validation set and threshold estimation on the test set. It also outperforms models selected using class-level labels, and then dynamically thresholded with only LOC maps.

Leveraging spatiotemporal information in videos is critical for weakly supervised video object localization (WSVOL) tasks. However, state-of-the-art methods only rely on visual and motion cues, while discarding discriminative information, making them susceptible to inaccurate localizations. Recently, discriminative models have been explored for WSVOL tasks using a temporal class activation mapping (CAM) method. Although their results are promising, objects are assumed to have limited movement from frame to frame, leading to degradation in performance for relatively long-term dependencies. This paper proposes a novel CAM method for WSVOL that exploits spatiotemporal information in activation maps during training without constraining an object’s position. Its training relies on Co - Lo calization, hence, the name CoLo-CAM . Given a sequence of frames, localization is jointly learned based on color cues extracted across the corresponding maps, by assuming that an object has similar color in consecutive frames. CAM activations are constrained to respond similarly over pixels with similar colors, achieving co-localization. This improves localization performance because the joint learning creates direct communication among pixels across all image locations and over all frames, allowing for transfer, aggregation, and correction of localizations. Co-localization is integrated into training by minimizing the color term of a conditional random field (CRF) loss over a sequence of frames/CAMs. Extensive experiments 1 on two challenging YouTube-Objects datasets of unconstrained videos show the merits of our CoLo-CAM method, and its robustness to long-term dependencies, leading to new state-of-the-art performance for WSVOL task.

Deep learning models for multimodal expression recognition have reached remarkable performance in controlled laboratory environments because of their ability to learn complementary and redundant semantic information. However, these models struggle in the wild, mainly because of the unavailability and quality of modalities used for training. In practice, only a subset of the training-time modalities may be available at test time. Learning with privileged information enables models to exploit data from additional modalities that are only available during training. State-of-the-art knowledge distillation (KD) methods have been proposed to distill information from multiple teacher models (each trained on a modality) to a common student model. These privileged KD methods typically utilize point-to-point matching, yet have no explicit mechanism to capture the structural information in the teacher representation space formed by introducing the privileged modality. We argue that encoding this same structure in the student space may lead to enhanced student performance. This paper introduces a new structural KD mechanism based on optimal transport (OT), where entropy-regularized OT distills the structural dark knowledge. Our privileged KD with OT (PKDOT) method captures the local structures in the multimodal teacher representation by calculating a cosine similarity matrix and selecting the top-k anchors to allow for sparse OT solutions, resulting in a more stable distillation process. Experiments1were performed on two challenging problems - pain estimation on the Biovid dataset (ordinal classification) and arousal-valance prediction on the Affwild2 dataset (regression). Results show that our proposed method can outperform state-of-the-art privileged KD methods on these problems. The diversity among modalities and fusion architectures indicates that PKDOT is modality-and model-agnostic.

Although state-of-the-art classifiers for facial expression recognition (FER) can achieve a high level of accuracy, they lack interpretability, an important feature for end-users. Experts typically associate spatial action units (AUs) from a codebook to facial regions for the visual interpretation of expressions. In this paper, the same expert steps are followed. A new learning strategy is proposed to explicitly incorporate AU cues into classifier training, allowing to train deep interpretable models. During training, this AU codebook is used, along with the input image expression label, and facial landmarks, to construct a AU heatmap that indicates the most discriminative image regions of interest w.r.t the facial expression. This valuable spatial cue is leveraged to train a deep interpretable classifier for FER. This is achieved by constraining the spatial layer features of a classifier to be correlated with AU heatmaps. Using a composite loss, the classifier is trained to correctly classify an image while yielding interpretable visual layer-wise attention correlated with AU maps, simulating the expert decision process. Our strategy only relies on image class expression for supervision, without additional manual annotations. Our new strategy is generic, and can be applied to any deep CNN - or transformer-based classifier without requiring any architectural change or significant additional training time. Our extensive evaluation11Our code is available at:https://github.com/sbelharbi/interpretable-fer-aus. on two public benchmarks RAF-DB, and AffectNet datasets shows that our proposed strategy can improve layer-wise interpretability without degrading classification performance. In addition, we explore a common type of interpretable classifiers that rely on class activation mapping (CAM) methods, and show that our approach can also improve CAM interpretability.

Adapting a deep learning model to a specific target individual is a challenging facial expression recognition (FER) task that may be achieved using unsupervised domain adaptation (UDA) methods. Although several UDA methods have been proposed to adapt deep FER models across source and target data sets, multiple subject-specific source domains are needed to accurately represent the intra-and inter-person variability in subject-based adaption. This paper considers the setting where domains correspond to individuals, not entire datasets. Unlike UDA, multi-source domain adaptation (MSDA) methods can leverage multiple source datasets to improve the accuracy and robustness of the target model. However, previous methods for MSDA adapt image classification models across datasets and do not scale well to a more significant number of source domains. This paper introduces a new MSDA method for subject-based domain adaptation in FER. It efficiently leverages information from multiple source subjects (labeled source domain data) to adapt a deep FER model to a single target individual (unlabeled target domain data). During adaptation, our subject-based MSDA first computes a between-source discrepancy loss to mitigate the domain shift among data from several source subjects. Then, a new strategy is employed to generate augmented confident pseudo-labels for the target subject, allowing a reduction in the domain shift between source and target subjects. Experiments1performed on the challenging BioVid heat and pain dataset with 87 subjects and the UNBC-McMaster shoulder pain dataset with 25 subjects show that our subject-based MSDA can outperform state-of-the-art methods yet scale well to multiple subject-based source domains.

Unsupervised object discovery is becoming an essential line of research for tackling recognition problems that require decomposing an image into entities, such as semantic segmentation and object detection. Recently, object-centric methods that leverage self-supervision have gained popularity, due to their simplicity and adaptability to different settings and conditions. However, those methods do not exploit effective techniques already employed in modern self-supervised approaches. In this work, we consider an object-centric approach in which DINO ViT features are reconstructed via a set of queried representations called slots. Based on that, we propose a masking scheme on input features that selectively disregards the background regions, inducing our model to focus more on salient objects during the reconstruction phase. Moreover, we extend the slot attention to a multi-query approach, allowing the model to learn multiple sets of slots, producing more stable masks. During training, these multiple sets of slots are learned independently while, at test time, these sets are merged through Hungarian matching to obtain the final slots. Our experimental results and ablations on the PASCAL-VOC 2012 dataset show the importance of each component and highlight how their combination consistently improves object localization. Our source code is available at: github.com/rishavpramanik/maskedmultiqueryslot

Data augmentation is one of the most effective techniques for regularizing deep learning models and improving recognition performance in various tasks and domains. However, this holds for standard in-domain settings, in which the training and test data follow the same distribution. For the out-of-domain case, where the test data follow a different and unknown distribution, the best recipe for data augmentation is unclear. In this paper, we show that for out-of-domain and domain generalization settings, data augmentation can provide a conspicuous and robust improvement in performance. To do that, we propose a simple training procedure: (i) use uniform sampling on standard data augmentation transformations; (ii) increase the strength transformations to account for the higher data variance expected when working out-of-domain, and (iii) devise a new reward function to reject extreme transformations that can harm the training. With this procedure, our data augmentation scheme achieves a level of accuracy comparable to or better than state-of-the-art methods on benchmark domain generalization datasets. Code: https://github.com/Masseeh/DCAug

Adapting visual object detectors to operational target domains is a challenging task, commonly achieved using unsupervised domain adaptation (UDA) methods. Recent studies have shown that when the labeled dataset comes from multiple source domains, treating them as separate domains and performing a multi-source domain adaptation (MSDA) improves the accuracy and robustness over blending these source domains and performing a UDA. For adaptation, existing MSDA methods learn domain-invariant and domain-specific parameters (for each source domain). However, unlike single-source UDA methods, learning domain-specific parameters makes them grow significantly in proportion to the number of source domains. This paper proposes a novel MSDA method called Prototype-based Mean Teacher (PMT), which uses class prototypes instead of domain-specific subnets to encode domain-specific information. These prototypes are learned using a contrastive loss, aligning the same categories across domains and separating different categories far apart. Given the use of prototypes, the number of parameters required for our PMT method does not increase significantly with the number of source domains, thus reducing memory issues and possible overfitting. Empirical studies indicate that PMT outperforms state-of-the-art MSDA methods on several challenging object detection datasets. Our code is available at https://github.com/imatif17/Prototype-Mean-Teacher

A powerful way to adapt a visual recognition model to a new domain is through image translation. However, common image translation approaches only focus on generating data from the same distribution as the target domain. Given a cross-modal application, such as pedestrian detection from aerial images, with a considerable shift in data distribution between infrared (IR) to visible (RGB) images, a translation focused on generation might lead to poor performance as the loss focuses on irrelevant details for the task. In this paper, we propose HalluciDet, an IR-RGB image translation model for object detection. Instead of focusing on reconstructing the original image on the IR modality, it seeks to reduce the detection loss of an RGB detector, and therefore avoids the need to access RGB data. This model produces a new image representation that enhances objects of interest in the scene and greatly improves detection performance. We empirically compare our approach against state-of-the-art methods for image translation and for fine-tuning on IR, and show that our HalluciDet improves detection accuracy in most cases by exploiting the privileged information encoded in a pre-trained RGB detector. Code: https://github.com/heitorrapela/HalluciDet.

Fully Test-Time Adaptation (TTA), which aims at adapting models to data drifts, has recently attracted wide interest. Numerous tricks and techniques have been proposed to ensure robust learning on arbitrary streams of unlabeled data. However, assessing the true impact of each individual technique and obtaining a fair comparison still constitutes a significant challenge. To help consolidate the community’s knowledge, we present a categorization of selected orthogonal TTA techniques, including small batch normalization, stream rebalancing, reliable sample selection, and network confidence calibration. We meticulously dissect the effect of each approach on different scenarios of interest. Through our analysis, we shed light on trade-offs induced by those techniques between accuracy, the computational power required, and model complexity. We also uncover the synergy that arises when combining techniques and are able to establish new state-of-the-art results.

The recent emergence of new algorithms for permuting models into functionally equivalent regions of the solution space has shed some light on the complexity of error surfaces and some promising properties like mode connectivity. However, finding the permutation that minimizes some ob-jectives is challenging, and current optimization techniques are not differentiable, which makes it difficult to integrate into a gradient-based optimization, and often leads to sub-optimal solutions. In this paper, we propose a Sinkhorn re-basin network with the ability to obtain the transportation plan that better suits a given objective. Unlike the current state-of-art, our method is differentiable and, there-fore, easy to adapt to any task within the deep learning do-main. Furthermore, we show the advantage of our re-basin method by proposing a new cost function that allows per-forming incremental learning by exploiting the linear mode connectivity property. The benefit of our method is compared against similar approaches from the literature under several conditions for both optimal transport and linear mode connectivity. The effectiveness of our continual learning method based on re-basin is also shown for several common benchmark datasets, providing experimental results that are competitive with the state-of-art. The source code is provided at https://github.com/jagp/sinkhorn-rebasin.

Systems for person re-identification (ReID) can achieve a high accuracy when trained on large fully-labeled image datasets. However, the domain shift typically associated with diverse operational capture conditions (e.g., camera viewpoints and lighting) may translate to a significant decline in performance. This paper focuses on unsupervised domain adaptation (UDA) for video-based ReID – a relevant scenario that is less explored in the literature. In this scenario, the ReID model must adapt to a complex target domain defined by a network of diverse video cameras based on track-let information. State-of-art methods cluster unlabeled target data, yet domain shifts across target cameras (sub-domains) can lead to poor initialization of clustering methods that propagates noise across epochs, thus preventing the ReID model to accurately associate samples of same identity. In this paper, an UDA method is introduced for video person ReID that leverages knowledge on video tracklets, and on the distribution of frames captured over target cameras to improve the performance of CNN backbones trained using pseudo-labels. Our method relies on an adversarial approach, where a camera-discriminator network is introduced to extract discriminant camera-independent representations, facilitating the subsequent clustering. In addition, a weighted contrastive loss is proposed to leverage the confidence of clusters, and mitigate the risk of incorrect identity associations. Experimental results obtained on three challenging video-based person ReID datasets – PRID2011, iLIDS-VID, and MARS – indicate that our proposed method can outperform related state-of-the-art methods. Our code is available at: https://github.com/dmekhazni/CAWCL-ReID

Synthetic image generation has recently experienced significant improvements in domains such as natural image or art generation. However, the problem of figure and diagram generation remains unexplored. A challenging aspect of generating figures and diagrams is effectively rendering readable texts within the images. To alleviate this problem, we present OCR-VQGAN, an image encoder, and decoder that leverages OCR pre-trained features to optimize a text perceptual loss, encouraging the architecture to preserve high-fidelity text and diagram structure. To explore our approach, we introduce the Paper2Fig100k dataset, with over 100k images of figures and texts from research papers. The figures show architecture diagrams and methodologies of articles available at arXiv.org from fields like artificial intelligence and computer vision. Figures usually include text and discrete objects, e.g., boxes in a diagram, with lines and arrows that connect them. We demonstrate the effectiveness of OCR-VQGAN by conducting several experiments on the task of figure reconstruction. Additionally, we explore the qualitative and quantitative impact of weighting different perceptual metrics in the overall loss function. We release code, models, and dataset at https://github.com/joanrod/ocr-vqgan.

Deep learning models for semi-supervised medical image segmentation have achieved unprecedented performance for a wide range of tasks. Despite their high accuracy, these models may however yield predictions that are considered anatomically impossible by clinicians. Moreover, incorporating complex anatomical constraints into standard deep learning frameworks remains challenging due to their non-differentiable nature. To address these limitations, we propose a Constrained Adversarial Training (CAT) method that learns how to produce anatomically plausible segmentations. Unlike approaches focusing solely on accuracy measures like Dice, our method considers complex anatomical constraints like connectivity, convexity, and symmetry which cannot be easily modeled in a loss function. The problem of non-differentiable constraints is solved using a Reinforce algorithm which enables to obtain a gradient for violated constraints. To generate constraint-violating examples on the fly, and thereby obtain useful gradients, our method adopts an adversarial training strategy which modifies training images to maximize the constraint loss, and then updates the network to be robust to these adversarial examples. The proposed method offers a generic and efficient way to add complex segmentation constraints on top of any segmentation network. Experiments on synthetic data and four clinically-relevant datasets demonstrate the effectiveness of our method in terms of segmentation accuracy and anatomical plausibility.

We present an unsupervised domain adaptation method for image segmentation which aligns high-order statistics, computed for the source and target domains, encoding domain-invariant spatial relationships between segmentation classes. Our method first estimates the joint distribution of predictions for pairs of pixels whose relative position corresponds to a given spatial displacement. Domain adaptation is then achieved by aligning the joint distributions of source and target images, computed for a set of displacements. Two enhancements of this method are proposed. The first one uses an efficient multi-scale strategy that enables capturing long-range relationships in the statistics. The second one extends the joint distribution alignment loss to features in intermediate layers of the network by computing their cross-correlation. We test our method on the task of unpaired multi-modal cardiac segmentation using the Multi-Modality Whole Heart Segmentation Challenge dataset and prostate segmentation task where images from two datasets are taken as data in different domains. Our results show the advantages of our method compared to recent approaches for cross-domain image segmentation. Code is available at https://github.com/WangPing521/Domain_adaptation_shape_prior.

Semi- and weakly-supervised learning have recently attracted considerable attention in the object detection literature since they can alleviate the cost of annotation needed to successfully train deep learning models. State-of-art approaches for semi-supervised learning rely on student-teacher models trained using a multi-stage process, and considerable data augmentation. Custom networks have been developed for the weakly-supervised setting, making it difficult to adapt to different detectors. In this paper, a weakly semi-supervised training method is introduced that reduces these training challenges, yet achieves state-of-the-art performance by leveraging only a small fraction of fully-labeled images with information in weakly-labeled images. In particular, our generic sampling-based learning strategy produces pseudo ground-truth (GT) bounding box annotations in an online fashion, eliminating the need for multi-stage training, and student-teacher network configurations. These pseudo GT boxes are sampled from weakly-labeled images based on the categorical score of object proposals accumulated via a score propagation process. Empirical results11Our code is available at: https://github.com/akhilpm/SemiWSOD on the Pascal VOC dataset, indicates that the proposed approach improves performance by 5.0% when using VOC 2007 as fully-labeled, and VOC 2012 as weak-labeled data. Also, with 5-10% fully annotated images, we observed an improvement of more than 10% in$m$AP, showing that a modest investment in image-level annotation, can substantially improve detection performance.

Class Activation Mapping (CAM) methods have recently gained much attention for weakly-supervised object localization (WSOL) tasks. They allow for CNN visualization and interpretation without training on fully annotated image datasets. CAM methods are typically integrated within off-the-shelf CNN backbones, such as ResNet50. Due to convolution and pooling operations, these backbones yield low resolution CAMs with a down-scaling factor of up to 32, contributing to inaccurate localizations. Interpolation is required to restore full size CAMs, yet it does not consider the statistical properties of objects, such as color and texture, leading to activations with inconsistent boundaries, and inaccurate localizations. As an alternative, we introduce a generic method for parametric upscaling of CAMs that allows constructing accurate full resolution CAMs (FCAMs). In particular, we propose a trainable decoding architecture that can be connected to any CNN classifier to produce highly accurate CAM localizations. Given an original low resolution CAM, foreground and background pixels are randomly sampled to fine-tune the decoder. Additional priors such as image statistics and size constraints are also considered to expand and refine object boundaries. Extensive experiments1, over three CNN backbones and six WSOL baselines on the CUB-200-2011 and OpenImages datasets, indicate that our F-CAM method yields a significant improvement in CAM localization accuracy. F-CAM performance is competitive with state-of-art WSOL methods, yet it requires fewer computations during inference.

Adversarial attack algorithms are dominated by penalty methods, which are slow in practice, or more efficient distance-customized methods, which are heavily tailored to the properties of the distance considered. We propose a white-box attack algorithm to generate minimally perturbed adversarial examples based on Augmented Lagrangian principles. We bring several algorithmic modifications, which have a crucial effect on performance. Our attack enjoys the generality of penalty methods and the computational efficiency of distance-customized algorithms, and can be readily used for a wide set of distances. We compare our attack to state-of-the-art methods on three datasets and several models, and consistently obtain competitive performances with similar or lower computational complexity.

The contrastive pre-training of a recognition model on a large dataset of unlabeled data often boosts the model’s performance on downstream tasks like image classification. However, in domains such as medical imaging, collecting unlabeled data can be challenging and expensive. In this work, we consider the task of medical image segmentation and adapt contrastive learning with meta-label annotations to scenarios where no additional unlabeled data is available. Meta-labels, such as the location of a 2D slice in a 3D MRI scan, often come for free during the acquisition process. We use these meta-labels to pre-train the image encoder, as well as in a semi-supervised learning step that leverages a reduced set of annotated data. A self-paced learning strategy exploiting the weak annotations is proposed to furtherhelp the learning process and discriminate useful labels from noise. Results on five medical image segmentation datasets show that our approach: i) highly boosts the performance of a model trained on a few scans, ii) outperforms previous contrastive and semi-supervised approaches, and iii) reaches close to the performance of a model trained on the full data.

Training deep learning models for accurate spatiotemporal recognition of facial expressions in videos requires significant computational resources. For practical reasons, 3D Convolutional Neural Networks (3D CNNs) are usually trained with relatively short clips randomly extracted from videos. However, such uniform sampling is generally sub-optimal because equal importance is assigned to each temporal clip. In this paper, we present a strategy for efficient video-based training of 3D CNNs. It relies on softmax temporal pooling and a weighted sampling mechanism to select the most relevant training clips. The proposed softmax strategy provides several advantages – a reduced computational complexity due to efficient clip sampling, and an improved accuracy since temporal weighting focuses on more relevant clips during both training and inference. Experimental results obtained with the proposed method on several facial expression recognition benchmarks show the benefits of focusing on more informative clips in training videos. In particular, our approach improves performance and computational cost by reducing the impact of inaccurate trimming and coarse annotation of videos, and heterogeneous distribution of visual information across time.

Despite the initial belief that Convolutional Neural Networks (CNNs) are driven by shapes to perform visual recognition tasks, recent evidence suggests that texture bias in CNNs provides higher performing models when learning on large labeled training datasets. This contrasts with the perceptual bias in the human visual cortex, which has a stronger preference towards shape components. Perceptual differences may explain why CNNs achieve human-level performance when large labeled datasets are available, but their performance significantly degrades in low-labeled data scenarios, such as few-shot semantic segmentation. To remove the texture bias in the context of few-shot learning, we propose a novel architecture that integrates a set of Difference of Gaussians (DoG) to attenuate high-frequency local components in the feature space. This produces a set of modified feature maps, whose high-frequency components are diminished at different standard deviation values of the Gaussian distribution in the spatial domain. As this results in multiple feature maps for a single image, we employ a bi-directional convolutional long-short-term-memory to efficiently merge the multi scale-space representations. We perform extensive experiments on three well-known few-shot segmentation benchmarks -Pascal i5, COCO-20i and FSS-1000- and demonstrate that our method outperforms state-of-the-art approaches in two datasets under the same conditions.

Data augmentation is a key practice in machine learning for improving generalization performance. However, finding the best data augmentation hyperparameters requires domain knowledge or a computationally demanding search. We address this issue by proposing an efficient approach to automatically train a network that learns an effective distribution of transformations to improve its generalization. Using bilevel optimization, we directly optimize the data augmentation parameters using a validation set. This framework can be used as a general solution to learn the optimal data augmentation jointly with an end task model like a classifier. Results show that our joint training method produces an image classification accuracy that is comparable to or better than carefully hand-crafted data augmentation. Yet, it does not need an expensive external validation loop on the data augmentation hyperparameters.

In the context of recent deep clustering studies, discriminative models dominate the literature and report the most competitive performances. These models learn a deep discriminative neural network classifier in which the labels are latent. Typically, they use multinomial logistic regression posteriors and parameter regularization, as is very common in supervised learning. It is generally acknowledged that discriminative objective functions (e.g., those based on the mutual information or the KL divergence) are more flexible than generative approaches (e.g., K-means) in the sense that they make fewer assumptions about the data distributions and, typically, yield much better unsupervised deep learning results. On the surface, several recent discriminative models may seem unrelated to K-means. This study shows that these models are, in fact, equivalent to K-means under mild conditions and common posterior models and parameter regularization. We prove that, for the commonly used logistic regression posteriors, maximizing the L2L2 regularized mutual information via an approximate alternating direction method (ADM) is equivalent to minimizing a soft and regularized K-means loss. Our theoretical analysis not only connects directly several recent state-of-the-art discriminative models to K-means, but also leads to a new soft and regularized deep K-means algorithm, which yields competitive performance on several image clustering benchmarks.

Weakly-supervised object localization is a challenging task in which the object of interest should be localized while learning its appearance. State-of-the-art methods recycle the architecture of a standard CNN by using the activation maps of the last layer for localizing the object. While this approach is simple and works relatively well, object localization relies on different features than classification, thus, a specialized localization mechanism is required during training to improve performance. In this paper, we propose a convolutional, multi-scale spatial localization network that provides accurate localization for the object of interest. Experimental results on CUB-200-2011 and ImageNet datasets show that our proposed approach provides competitive performance for weakly supervised localization.

Although deep neural networks (NNs) have achieved state-of-the-art accuracy in many visual recognition tasks, the growing computational complexity and energy consumption of networks remains an issue, especially for applications on platforms with limited resources and requiring real-time processing. Filter pruning techniques have recently shown promising results for the compression and acceleration of convolutional NNs (CNNs). However, these techniques involve numerous steps and complex optimisations because some only prune after training CNNs, while others prune from scratch during training, by integrating sparsity constraints or by modifying the loss function. In this paper, we introduce a new Progressive Gradient Pruning (PGP) technique for iterative filter pruning during training. In contrast to previous progressive pruning techniques, it relies on a novel filter selection criterion that measures the change in filter weights, uses a new hard and soft pruning strategy, and effectively adapts momentum tensors during the backward propagation pass. Experimental results obtained after training various CNNs on benchmark datasets for image classification, object detection and domain adaptation indicate that our PGP technique can achieve a better trade-off between classification accuracy and network (time and memory) complexity than PSFP and other state-of-the-art filter pruning techniques. Code is available on GitHub link: https://github.com/Anon6627/Pruning-PGP.

In this paper we propose a new approach for classifying the global emotion of images containing groups of people. To achieve this task, we consider two different and complementary sources of information: i) a global representation of the entire image (ii) a local representation where only faces are considered. While the global representation of the image is learned with a convolutional neural network (CNN), the local representation is obtained by merging face features through an attention mechanism. The two representations are first learned independently with two separate CNN branches and then fused through concatenation in order to obtain the final group-emotion classifier. For our submission to the EmotiW 2018 group-level emotion recognition challenge, we combine several variations of the proposed model into an ensemble, obtaining a final accuracy of 64.83% on the test set and ranking 4th among all challenge participants.

We propose “Areas of Attention”, a novel attentionbased model for automatic image captioning. Our approach models the dependencies between image regions, caption words, and the state of an RNN language model, using three pairwise interactions. In contrast to previous attentionbased approaches that associate image regions only to the RNN state, our method allows a direct association between caption words and image regions. During training these associations are inferred from image-level captions, akin to weakly-supervised object detector training. These associations help to improve captioning by localizing the corresponding regions during testing. We also propose and compare different ways of generating attention areas: CNN activation grids, object proposals, and spatial transformers nets applied in a convolutional fashion. Spatial transformers give the best results. They allow for image specific attention areas, and can be trained jointly with the rest of the network. Our attention mechanism and spatial transformer attention areas together yield state-of-the-art results on the MSCOCO dataset.

In this paper, a new method for generating object and action proposals in images and videos is proposed. It builds on activations of different convolutional layers of a pretrained CNN, combining the localization accuracy of the early layers with the high informativeness (and hence recall) of the later layers. To this end, we build an inverse cascade that, going backward from the later to the earlier convolutional layers of the CNN, selects the most promising locations and refines them in a coarse-to-fine manner. The method is efficient, because (i) it re-uses the same features extracted for detection, (ii) it aggregates features using integral images, and (iii) it avoids a dense evaluation of the proposals thanks to the use of the inverse coarse-to-fine cascade. The method is also accurate. We show that DeepProposals outperform most of the previous object proposal and action proposal approaches and, when plugged into a CNN-based object detector, produce state-of-the-art detection performance.

Weakly supervised object detection, is a challenging task, where the training procedure involves learning at the same time both, the model appearance and the object location in each image. The classical approach to solve this problem is to consider the location of the object of interest in each image as a latent variable and minimize the loss generated by such latent variable during learning. However, as learning appearance and localization are two interconnected tasks, the optimization is not convex and the procedure can easily get stuck in a poor local minimum, i.e. the algorithm “misses” the object in some images. In this paper, we help the optimization to get close to the global minimum by enforcing a “soft” similarity between each possible location in the image and a reduced set of “exemplars”, or clusters, learned with a convex formulation in the training images. The help is effective because it comes from a different and smooth source of information that is not directly connected with the main task. Results show that our method improves a strong baseline based on convolutional neural network features by more than 4 points without any additional features or extra computation at testing time but only adding a small increment of the training time due to the convex clustering.

In this paper we evaluate the quality of the activation layers of a convolutional neural network (CNN) for the generation of object proposals. We generate hypotheses in a sliding-window fashion over different activation layers and show that the final convolutional layers can find the object of interest with high recall but poor localization due to the coarseness of the feature maps. Instead, the first layers of the network can better localize the object of interest but with a reduced recall. Based on this observation we design a method for proposing object locations that is based on CNN features and that combines the best of both worlds. We build an inverse cascade that, going from the final to the initial convolutional layers of the CNN, selects the most promising object locations and refines their boxes in a coarse-to-fine manner. The method is efficient, because i) it uses the same features extracted for detection, ii) it aggregates features using integral images, and iii) it avoids a dense evaluation of the proposals due to the inverse coarse-to-fine cascade. The method is also accurate, it outperforms most of the previously proposed object proposals approaches and when plugged into a CNN-based detector produces state-of-the-art detection performance.

A common issue in deformable object detection is finding a good way to position the parts. This issue is even more outspoken when considering detection and pose estimation for 3D objects, where parts should be placed in a three-dimensional space. Some methods extract the 3D shape of the object from 3D CAD models. This limits their applicability to categories for which such models are available. Others represent the object with a predefined and simple shape (e.g. a cuboid). This extends the applicability of the model, but in many cases the pre-defined shape is too simple to properly represent the object in 3D. In this paper we propose a new method for the detection and pose estimation of 3D objects, that does not use any 3D CAD model or other 3D information. Starting from a simple and general 3D shape, we learn in a weakly supervised manner the 3D part locations that best fit the training data. As this method builds on a iterative estimation of the part locations, we introduce several speedups to make the method fast enough for practical experiments. We evaluate our model for the detection and pose estimation of faces and cars. Our method obtains results comparable with the state of the art, it is faster than most of the other approaches and does not need any additional 3D information.

Query-by-example remains popular in image retrieval because it can exploit contextual information encoded in the image, that is difficult to express in a traditional textual query. Textual queries, on the other hand, give more flexibility in that it's easy to reformulate and refine a text query based on initial results.

In this paper we present a scalable way to learn and detect objects using a 3D representation based on HOG patches placed on a 3D cuboid. The model consists of a single 3D representation that is shared among views. Similarly to the work of Fidler et al. [5], at detection time this representation is projected on the image plane over the desired viewpoints. However, whereas in [5] the projection is done at image-level and therefore the computational cost is linear in the number of views, in our model every view is approximated at feature level as a linear combination of the pre-computed fron to-parallel views. As a result, once the fron to-parallel views have been computed, the cost of computing new views is almost negligible. This allows the model to be evaluated on many more viewpoints. In the experimental results we show that the proposed model has a comparable detection and pose estimation performance to standard multiview HOG detectors, but it is faster, it scales very well with the number of views and can better generalize to unseen views. Finally, we also show that with a procedure similar to label propagation it is possible to train the model even without using pose annotations at training time.

In classification of objects substantial work has gone into improving the low level representation of an image by considering various aspects such as different features, a number of feature pooling and coding techniques and considering different kernels. Unlike these works, in this paper, we propose to enhance the semantic representation of an image. We aim to learn the most important visual components of an image and how they interact in order to classify the objects correctly. To achieve our objective, we propose a new latent SVM model for category level object classification. Starting from image-level annotations, we jointly learn the object class and its context in terms of spatial location (where) and appearance (what). Furthermore, to regularize the complexity of the model we learn the spatial and co-occurrence relations between adjacent regions, such that unlikely configurations are penalized. Experimental results demonstrate that the proposed method can consistently enhance results on the challenging Pascal VOC dataset in terms of classification and weakly supervised detection. We also show how semantic representation can be exploited for finding similar content.

Face detection and facial points localization are interconnected tasks. Recently it has been shown that solving these two tasks jointly with a mixture of trees of parts (MTP) leads to state-of-the-art results. However, MTP, as most other methods for facial point localization proposed so far, requires a complete annotation of the training data at facial point level. This is used to predefine the structure of the trees and to place the parts correctly. In this work we extend the mixtures from trees to more general loopy graphs. In this way we can learn in a weakly supervised manner (using only the face location and orientation) a powerful deformable detector that implicitly aligns its parts to the detected face in the image. By attaching some reference points to the correct parts of our detector we can then localize the facial points. In terms of detection our method clearly outperforms the state-of-the-art, even if competing with methods that use facial point annotations during training. Additionally, without any facial point annotation at the level of individual training images, our method can localize facial points with an accuracy similar to fully supervised approaches.

Recently a lot of progress has been made in the field of video segmentation. The question then arises whether and how these results can be exploited for this other video processing challenge, action recognition. In this paper we show that a good segmentation is actually very important for recognition. We propose and evaluate several ways to integrate and combine the two tasks: i) recognition using a standard, bottom-up segmentation, ii) using a top-down segmentation geared towards actions, iii) using a segmentation based on inter-video similarities (co-segmentation), and iv) tight integration of recognition and segmentation via iterative learning. Our results clearly show that, on the one hand, the two tasks are interdependent and therefore an iterative optimization of the two makes sense and gives better results. On the other hand, comparable results can also be obtained with two separate steps but mapping the feature-space with a non-linear kernel.

Most advanced driving assistance systems already include pedestrian detection systems. Unfortunately, there is still a tradeoff between precision and real time. For a reliable detection, excellent precision-recall such a tradeoff is needed to detect as many pedestrians as possible while, at the same time, avoiding too many false alarms; in addition, a very fast computation is needed for fast reactions to dangerous situations. Recently, novel approaches based on deformable templates have been proposed since these show a reasonable detection performance although they are computationally too expensive for real-time performance. In this paper, we present a system for pedestrian detection based on a hierarchical multiresolution part-based model. The proposed system is able to achieve state-of-the-art detection accuracy due to the local deformations of the parts while exhibiting a speedup of more than one order of magnitude due to a fast coarse-to-fine inference technique. Moreover, our system explicitly infers the level of resolution available so that the detection of small examples is feasible with a very reduced computational cost. We conclude this contribution by presenting how a graphics processing unit-optimized implementation of our proposed system is suitable for real-time pedestrian detection in terms of both accuracy and speed.

We present a method that can dramatically accelerate object detection with part based models. The method is based on the observation that the cost of detection is likely to be dominated by the cost of matching each part to the image, and not by the cost of computing the optimal configuration of the parts as commonly assumed. Therefore accelerating detection requires minimizing the number of part-to-image comparisons. To this end we propose a multiple-resolutions hierarchical part based model and a corresponding coarse-to-fine inference procedure that recursively eliminates from the search space unpromising part placements. The method yields a ten-fold speedup over the standard dynamic programming approach and is complementary to the cascade-of-parts approach of. Compared to the latter, our method does not have parameters to be determined empirically, which simplifies its use during the training of the model. Most importantly, the two techniques can be combined to obtain a very significant speedup, of two orders of magnitude in some cases. We evaluate our method extensively on the PASCAL VOC and INRIA datasets, demonstrating a very high increase in the detection speed with little degradation of the accuracy.